VIP Failover Time for SAN/IQ v8

[Paul Miano]

Has anyone timed the VIP failover when booting the node that hold the VIP? I have not found anything in the documentation - maybe I missed it. We are planning an upgrade from 8.0 to 8.1 (and eventually the newest once avaialble) and am trying to determine if the SQL databases are going to complain. We currently have a 2 node 2120 G1 cluster with FoM running on ESXi local storage. We are running 2 ESXi hosts (v3.5u3). We've upgraded SAN/IQ in the past, but not with any SQL running on the cluster. As we continue to migrate from DAS this will be more of an issue. We've looked into the multi-site SAN setup, as we have 2 2120G2s to add to the cluster soon, but have not completely decided on that yet.

[jennylovejasmine]

any update for this?

[kghammond · Accepted solution]

This does not directly answer your question, but the recommended procedures are to always do maintenance on the node(s) that does not own the VIP first.

I believe when a node comes online it will cleanly take ownership of the VIP thus allowing you to then do maintenance on the other node that used to own the VIP.

Our experience is that about 95% of the time, when a node comes online it takes ownership of the VIP. A few times that did not happen and then we waited for the volumes to resync, rebooted the node and it took ownerhsip of the VIP on the next reboot.

This procedure will never have downtime of the VIP and such your applications will always have access to your LUNS. Assuming they are network RAID 10.

Kevin

[jennylovejasmine]

Thank you!